USDA Forest Service Proceedings – RMRS-P-56
49.
From Genes to Ecosystems: Measuring
Evolutionary Diversity and Community
Structure with Forest Inventory and Analysis
(FIA) Data
Kevin M. Potter1
Abstract: Forest genetic sustainability is an important component of forest health
because genetic diversity and evolutionary processes allow for the adaptation of species
and for the maintenance of ecosystem functionality and resilience. Phylogenetic
community analyses, a set of new statistical methods for describing the evolutionary
relationships among species, offer an innovative approach for assessing the health of
forest communities from an evolutionary perspective. Forest Inventory and Analysis data
are ideal for conducting phylogenetic community analyses for forest tree species at broad
scales. FIA data from 100,000 plots across the conterminous United States were used to
investigate the evolutionary characteristics of forest tree communities. This required
generating a phylogenetic “evolutionary tree” of the 311 forest tree species inventoried
by FIA, based on recent gene sequencing studies of several plant groups. Phylogenetic
diversity was quantified for each plot; this statistic sums the evolutionary age of the
species present, using the phylogenetic tree. This is a more meaningful way to quantify
biodiversity than species richness because, rather than weighting all species on a plot
equally regardless of relatedness, it measures their cumulative evolutionary age.
General patterns at the ecoregion section scale were similar between mean plot-level
species richness and phylogenetic diversity, but important differences also existed.
Additionally, the analyses described each plot’s phylogenetic community structure, or
whether the species were more clustered or dispersed across the 311-species reference
evolutionary tree than expected by chance. The most phylogenetically overdispersed
sections were located in the Interior West, while the most phylogenetically clustered
included several in the Upper Midwest, New England, California and the Southeastern
Coastal Plain. Communities that are phylogenetically clustered consist of more closely
related species and therefore may be more susceptible to threats such as pests and
climate change. Phylogenetically dispersed communities may be more resilient to these
pressures, because greater evolutionary diversity is expected to translate into a greater
likelihood that more species will be unaffected by, or will be adaptable to, environmental
changes.
Keywords: Biodiversity, evolutionary biology, landscape genetics, forest health
monitoring.
1
North Carolina State University; Department of Forestry and Environmental Resources; Forest
Health Monitoring Program; 3041 Cornwallis Road, Research Triangle Park, NC 27709;
kevinpotter@fs.fed.us.
In: McWilliams, Will; Moisen, Gretchen; Czaplewski, Ray, comps. 2009. 2008 Forest Inventory and Analysis (FIA)
Symposium; October 21-23, 2008: Park City, UT. Proc. RMRS-P-56CD. Fort Collins, CO: U.S. Department of
Agriculture, Forest Service, Rocky Mountain Research Station. 1 CD.
USDA Forest Service Proceedings – RMRS-P-56
49.
Introduction
The integration of three diverse scientific disciplines – landscape ecology,
conservation biology, and evolutionary genetics – can offer innovative and
valuable insights about the health of forested ecosystems, particularly at large
scales (figure 1).
Approaching conservation from a landscape ecology
perspective is vitally important because large-scale processes affect biodiversity
and because management decisions are often made at regional levels. At the same
time, it is imperative to consider genetic processes in a landscape ecology context
because ecological dynamics at a variety of scales can affect the evolution of
species. Understanding conservation biology in the light of evolutionary
processes, meanwhile, is necessary to measure and manage forest biodiversity.
Forest health monitoring work that integrates all three disciplines of landscape
ecology, conservation, and genetics is now possible because of the availability of
recent advancements in computing, statistical analysis, and molecular biology.
This work is particularly relevant to assessing forest genetic sustainability across
large scales using U.S. Forest Service Forest Inventory and Analysis (FIA) Phase
2 inventory data.
Figure 1: Relationships among landscape ecology, conservation biology and evolutionary genetics
within the context of monitoring forest health and sustainability.
2
USDA Forest Service Proceedings – RMRS-P-56
49.
Forest genetic sustainability is an important component of forest health
because genetic diversity and evolutionary processes allow for the adaptation of
species and for the maintenance of ecosystem functionality and resilience.
Genetic diversity within species, for example, reflects the recent and existing
integrity of evolutionary and ecological processes (Brown et al. 2000). At the
same time, existing evolutionary lineages will generate future biodiversity, and
are therefore a cornerstone of environmental health (Erwin 1991). The
importance of forest genetic diversity is reflected in the Criteria and Indicators for
the Conservation and Sustainable Management of Temperate and Boreal Forests,
which includes three indicators of genetic diversity under Criterion 1
(conservation of biological diversity):
(1) number and geographic distribution of forest associated species at risk of
losing genetic variation and locally adapted genotypes,
(2) population levels of selected representative forest associated species to
describe genetic diversity, and
(3) status of on-site and off-site efforts focused on conservation of genetic
diversity (Montréal Process Working Group 2008).
The first two of these indicators attempt to quantify the number and population
levels of individual species at risk of losing genetic variability, while the third
indicator seeks to evaluate the success gene conservation efforts for forest species,
presumably those at risk. All three are valuable measurements, but each is limited
to a handful of indicator species and therefore does not address the evolutionary
diversity present across entire forest communities. A new approach, however, can
now combine nationwide FIA forest inventory data, an ecoregion-scale
perspective, and recent advancements in gene sequencing and molecular
systematics to quantify evolutionary diversity for communities of species across
large spatial scales.
This approach, community phylogenetic analyses (Webb and others 2002),
offers the ability to synthesize evolutionary biology and landscape ecology in the
context of assessing forest community health, both in terms of biodiversity and
community resilience to stress. Community phylogenetic analyses are now
possible across broad taxonomic species groupings (forest trees, for example) and
across large spatial scales because of recent advances in the ability to create
robust phylogenies describing the relationships among species. These
phylogenies are hypothesized “family trees” of species (figure 2) that are
developed by surveying molecular systematic studies, which compare differences
and similarities in gene sequences to determine the evolutionary relationships
among the species and to estimate how long ago species diverged from each
other. Fossil evidence from palaeobotanical studies can further calibrate the dates
at which species or species groups diverged. The branches within these
phylogenetic trees often are measured in millions of years. Constructing such
3
USDA Forest Service Proceedings – RMRS-P-56
49.
phylogenetic trees does not always require new gene sequencing work, because
molecular systematic studies now have been published for a wide variety of plant
groups.
Figure 2: A phylogenetic tree of the hypothesized evolutionary relationships among true fir (Abies)
species of North America inventoried by FIA. The scale bar at the bottom of the figure depicts time
in millions of years. Such phylogenetic trees are hypotheses of the relationships among species,
and can be improved over time with additional gene sequencing studies and with additional
information unearthed in the fossil record.
Such phylogenetic trees are necessary for two types of phylogenetic
community analyses that are useful in describing evolutionary relationships
within forest tree communities at broad spatial scales: the generation of
phylogenetic diversity statistics and the quantification of phylogenetic community
structure.
Phylogenetic diversity statistics (Faith 1992, Webb and others 2006) are
particularly meaningful as a measure of biodiversity because they quantify the
cumulative evolutionary age and evolutionary potential of all the species in the
community of interest. This is typically done by first generating a phylogenetic
tree encompassing the species present in a community, based on existing
molecular systematics research. Next, the lengths of the branches of that
phylogenetic tree, usually measured in millions of years of evolutionary time, are
measured and summed. This approach may be more meaningful than traditional
biodiversity metrics such as species richness and abundance, which weight the
value of all species equally regardless of their relatedness (figure 3). Maintaining
the evolutionary potential of groups of species measured in this way has become
an increasingly important conservation goal (Rodrigues and Gaston 2002,
Sechrest and others 2002, Soltis and Gitzendanner 1999).
4
USDA Forest Service Proceedings – RMRS-P-56
49.
Figure 3: A comparison between two measurements of biodiversity, species richness and
phylogenetic diversity. Species richness is the number of species present in the community, while
phylogenetic diversity is the summed evolutionary age of all the species in the community, as
measured by the phylogenetic tree encompassing the species present. Community (A) has greater
species richness than community (B), but community (B) has greater phylogenetic diversity
because the species in the community contain greater evolutionary distance. The branch lengths
shown on the phylogenetic trees are for demonstration purposes, and do not depict actual
evolutionary relationships among the species.
A second set of phylogenetic analyses focuses on determining whether the
species within a specific community are more phylogenetically clustered or
dispersed on the evolutionary tree of life than expected by chance (Webb 2000).
Such an analysis would, for example, compare the phylogenetic tree of the species
within a single FIA plot to the phylogenetic tree that includes all the tree species
in North America. If the species on the FIA plot are more closely related than
expected by random chance, when compared to the larger reference phylogenetic
tree, then the community is phylogenetically clustered (figure 4a). If the species
on the plot are less closely related than expected by chance, then the community
is phylogenetically overdispersed (figure 4b). This type of analysis is useful from
5
USDA Forest Service Proceedings – RMRS-P-56
49.
a forest health perspective because it offers a way to quantify a community’s
potential evolutionary resilience in the face of pests, pathogens, climate change
and other stresses. That is because communities made up of species that are
overdispersed on the phylogenetic tree possess greater-than-expected evolutionary
diversity, and may therefore encompass a higher proportion of species unaffected
by a given stressor or able to adapt to it. Phylogenetically clustered communities,
meanwhile, contain less evolutionary diversity and may be at greater cumulative
risk from stressors that might affect several closely related species, such as
sudden oak death (Phytophthora ramorum).
Figure 4: Phylogenetic trees of hypothetical tree communities representing the two types of
phylogenetic community structure, (A) phylogenetic clustering and (B) phylogenetic overdispersion.
The species in red are those present in the community; the species in black and red together
encompass the phylogenetic reference tree of species that could exist in the community.
The work described in this paper had three main objectives:
(1)
Assess the usefulness of FIA tree inventory datasets for phylogenetic
diversity and community structure analyses across large ecoregion
scales.
(2)
Compare mean FIA plot-level tree species richness to mean plot
phylogenetic diversity at the ecoregion section scale.
(3)
Test for correlations between mean plot measures of phylogenetic
community structure and environmental variables at the ecoregion
section scale.
6
USDA Forest Service Proceedings – RMRS-P-56
49.
Methods
The software package Phylocom 3.41 (Webb and others 2007) was used to
calculate phylogenetic diversity and phylogenetic community structure statistics
for the forest trees present on each of 102,304 one-sixth-acre Forest Inventory and
Analysis (FIA) plots across the 48 conterminous United States. These plots
represented the latest available FIA Phase 2 tree and sapling inventory data (trees
≥ 1 inch dbh) as of November 2007 (Forest Inventory and Analysis Program
2007). A phylogenetic reference supertree (figure 5), encompassing the 311
forest tree species inventoried by FIA across the conterminous United States, was
constructed for the estimation of phylogenetic distance among species in units of
millions of years (Potter in review).
Figure 5: The phylogenetic supertree used in this analysis, incorporating 311 North American
forest tree species inventoried by FIA. Evolutionary relationships and branch lengths were based
on a survey of approximately 70 recent molecular systematic and paleobotanical studies, with the
exception of basal angiosperm relationships, which taken from Wikstrom et al. (2001). Branch
lengths are measured in millions of years.
7
USDA Forest Service Proceedings – RMRS-P-56
49.
Based on its approximate geographic coordinates, each FIA plot was assigned
with ArcMap 9.2 (ESRI 2006) to the appropriate ecoregion section (Bailey 1995),
using the most recent geographical information system (GIS) mapping of Bailey’s
hierarchical system of ecoregion domains, divisions, provinces and sections
(Cleland and others 2005). Ecoregion sections containing fewer than 25 plots
were excluded from the analyses to ensure an adequate sampling.
Species richness was calculated for each FIA plot, as was Faith’s (1992) index
of phylogenetic diversity (PD). This quantifies the total evolutionary history
represented by the species in a community by determining the total phylogenetic
branch length distance of the species on the plot, divided by the total branch
length distance of the reference phylogenetic supertree encompassing all the
North American tree species inventoried by FIA. Mean PD and species richness
values were calculated for ecoregion sections.
Additionally, two measures of phylogenetic community structure were
calculated for each FIA plot. Both statistics measure whether the configuration of
the phylogenetic tree, encompassing the species in a given community, is more or
less clustered than expected by chance, as compared to a regional species pool. In
this analysis, the regional species pool for each plot consisted of all the species
inventoried on all the FIA plots within the same ecoregion section as the plot in
question. (See Potter (in review) for details about the calculation of these
statistics.)
(1)
The Nearest Taxon Index (NTI) is a standardized measure of the
branch-tip phylogenetic clustering or overdispersion of the species on
the FIA plot, regardless of the arrangement of the higher level groups
in the phylogenetic tree (Webb and others 2006). In other words, NTI
measures whether evolutionary diversity among the species on the plot
is higher (overdispersed) or lower (clustered) than expected by chance,
as compared to the ecoregion section species pool.
(2)
The Net Relatedness Index (NRI) is a standardized measurement of
basal or tree-wide phylogenetic clustering or overdispersion of the
species on the FIA plot (Webb and others 2006). In other words, NRI
measures whether evolutionary diversity among the deeper
phylogenetic ranks (families, orders, classes and divisions) present on
the plot is higher (overdispersed) or lower (clustered) than expected by
chance, again as compared to the ecoregion section species pool.
NTI and NRI measure different evolutionary characteristics of communities, so
the two statistics could show different results for a plot, for example indicating
clustering by one metric and overdispersion by the other. NTI and NRI values are
negative when species on a plot do not occur together with closely related species
(overdispersed), and positive when species occur with other closely related
species (clustered) (Kembel and Hubbell 2006). The means for plot-level NRI
8
USDA Forest Service Proceedings – RMRS-P-56
49.
and NTI values were calculated across ecoregion sections, with a t test
determining whether the section-level mean values were significantly different
from 0.
The possibility of large-scale correlations of these two phylogenetic
community structure measures with ecological variables was tested for several
climate (PRISM Group 2008), soil (Miller and White 1998) and topographic
(United States Geological Survey 1996) variables at the ecoregion section scale.
The ecoregion section means of each of these was determined across the forested
area of the section, using a forest cover map (1 km2 resolution) derived from
Moderate Resolution Imaging Spectroradiometer (MODIS) imagery by the U.S.
Forest Service Remote Sensing Applications Center (United States Department of
Agriculture Forest Service 2008).
Results and Discussion
Values of both mean plot species richness and phylogenetic diversity tended
to be higher in ecoregion sections of the eastern United States than in the western
part of the country (figure 6). The sections with the lowest means for both
statistics were located in the interior West. While the general pattern was similar
between these two measures of biodiversity, important differences were apparent.
Specifically, phylogenetic diversity was higher relative to species richness in
several ecoregion sections along the Pacific coast, from central California to
Washington, as well as in sections in the Northeast and in the Great Lakes region.
Not surprisingly, the 10 ecoregion sections with the highest mean species
richness values are all located in the southeastern United States. However, only
six of the 10 sections with the highest mean phylogenetic diversity are located in
the Southeast, with the other four in the Northeast. Ecoregion sections with high
phylogenetic diversity relative to species richness appear to have a combination of
both high angiosperm (flowering plant) and gymnosperm (cone-bearing) species
richness (Potter in review).
Meanwhile, the analysis of phylogenetic community structure detected
regional patterns in phylogenetic overdispersion and clustering (figure 7). Both
the branch-tip Nearest Taxon Index (NTI) and tree-wide Net Relatedness Index
(NRI) metrics indicated the existence of phylogenetic clustering in some of the
most northerly ecoregion sections, including those in the Northeast, the Great
Lakes region, the northern Great Plains, and the Pacific Northwest, as well as
along the Appalachian and Sierra Nevada mountain chains. Slight to moderate
overdispersion for both metrics was present for sections in the Interior West,
particularly in the Southwest.
9
USDA Forest Service Proceedings – RMRS-P-56
49.
A)
B)
Figure 6: Mean FIA plot-level forest tree species richness (A) and Faith’s index of phylogenetic
diversity (B) across ecoregion sections. The two statistics were divided into six equal interval
classes for comparison purposes.
At the same time, some differences were apparent between the two measures.
Perhaps the most notable difference is in the Southeast, where six sections were
overdispersed using NTI, which quantifies clustering at the tips of the
phylogenetic tree, but where none were overdispersed using the NRI, which
10
USDA Forest Service Proceedings – RMRS-P-56
49.
measures tree-wide clustering. In fact, four of these six sections were clustered
using the NRI. This is most likely the situation because gymnosperm diversity is
low in these sections, resulting in tree-wide phylogenetic clustering, while their
angiosperm diversity is moderate, with angiosperm species evenly spaced across
the phylogenetic tree on many plots (Potter in review).
Such differences may have important forest health implications.
A
community consisting of species that are overdispersed across the phylogenetic
tree but clustered toward the branch tips may contain fewer species susceptible to
a given threat. That community, however, may be more likely to lose important
ecological functions provided by any species that are eliminated. Meanwhile, a
community of species clustered across the phylogenetic tree but overdispersed at
the tips may better retain its ecological functionality in such a situation, but might
encompass more species at risk of elimination.
It is worth noting that it is possible for a community to possess high
phylogenetic diversity but to still have a clustered phylogenetic community
structure, and, as a result, to be more susceptible to certain threats. This is
because of important differences in the two metrics. Phylogenetic diversity
measures the total evolutionary diversity present in a community, while NRI and
NTI are indices that quantify the configuration of the evolutionary relatedness
among the species. Ecoregions section with high mean plot-level phylogenetic
diversity and a high degree of clustering include the Eastern Upper Peninsula
(212R), the Blue Ridge Mountains (M221D), the Catskill Mountains (211I), the
Maine-New Brunswick Foothills and Lowlands (211B), and the Klamath
Mountains (M261A).
The patterns of phylogenetic community structure detected in this study may
be driven at least in part by ecological interactions among species within the
community.
Phylogenetic clustering, for example, may be caused by
environmental filtering, the process that occurs when closely related species tend
to co-occur because they share similar tolerances to the abiotic environment
(Cavender-Bares and others 2004, Tofts and Silvertown 2000). Because the
species present in such communities share much evolutionary history and an
affinity for similar environmental conditions, they may be particularly susceptible
collectively to a certain threats, such as insects and diseases targeting specific
families or genera of tree species. Phylogenetic overdispersion in a community,
meanwhile, may indicate the existence of competitive exclusion. This occurs
when closely related species compete for similar environmental niches and
exclude each other from a community when they share limiting resources
(Cavender-Bares and others 2004, Tofts and Silvertown 2000). The ecological
integrity of such communities may be less at risk from changing conditions
because they encompass a wider variety of evolutionary adaptations to respond to
those changes.
11
USDA Forest Service Proceedings – RMRS-P-56
49.
A)
B)
Figure 7: Mean FIA plot-level measures of phylogenetic clustering across ecoregion sections,
using (A) the Nearest Taxon Index, a measure of clustering at the tips of the phylogenetic
branches, and (B) the Net Relatedness Index, a measure of clustering throughout the phylogenetic
tree. A t-test was used to determine whether the mean index values were significantly different from
0, with negative index values overdispersed and positive values clustered.
12
USDA Forest Service Proceedings – RMRS-P-56
49.
Several correlations were detected between mean phylogenetic community
structure values and environmental characteristics at the ecoregion section scale
(table 1). These correlations suggest that competitive exclusion for resources
among related species (phylogenetic overdispersion) is an important process in
ecoregion sections with less “hospitable” environments.
Meanwhile,
environmental filtering, which occurs when more closely related species tend to
occupy the same kinds of environments (phylogenetic clustering), may be more
common when environmental conditions are more hospitable, such as at sites
with moister and less compacted soils, at lower elevations, and with greater
amounts of precipitation.
For example, soil available water capacity was the environmental variable
with the strongest positive correlation phylogenetic clustering across ecoregion
sections, indicating communities with wetter soils were generally more
phylogenetically clustered. Greater soil acidity (lower pH) also was positively
associated with phylogenetic clustering, as was depth to bedrock. Among the
climate variables, annual precipitation had a weak but significant correlation with
NRI, indicating that forest tree communities in ecoregion sections with more
precipitation were more phylogenetically clustered using the phylogenetic treewide metric.
Standard deviation of monthly maximum and minimum
temperatures was negatively correlated with both NRI and NTI, suggesting that
variability in temperatures was associated with greater phylogenetic
overdispersion of communities in the ecoregion. Finally, three different
topographical measures were important in predicting phylogenetic structure, with
ecoregion section elevation mean, elevation variation (standard deviation) and
elevation range all correlated with greater phylogenetic overdispersion.
Table 1: Correlations between mean plot-level phylogenetic clustering statistics across
ecoregion sections and environmental variables at the section scale.
Environmental variable
NRI
NTI
Soils
Available water capacity (AWC)
0.475***
0.463***
Bulk density
-0.519***
-0.37***
pH
-0.355***
-0.249**
Depth to bedrock
0.34***
0.321***
Climate
Mean annual precipitation (1971-2000)
0.226**
ns
Monthly maximum temperature (1971-2000) (mean)
ns
ns
Monthly maximum temperature (1971-2000) (standard deviation)
-0.377***
-0.383***
Monthly minimum temperature (1971-2000) (mean)
ns
ns
Monthly minimum temperature (1971-2000) (standard deviation)
-0.345***
-0.347***
Topography
Elevation (mean)
-0.413***
-0.41***
Elevation (standard deviation)
-0.274***
-0.281***
Elevation (range)
-0.315***
-0.339***
*** p<0.001; ** p<0.01; ns, not significant
13
USDA Forest Service Proceedings – RMRS-P-56
49.
Conclusions and future work
The work outlined in this paper demonstrates the applicability of phylogenetic
community analysis to data sets at regional and even national scales. Generating
these phylogenetic diversity and community structure statistics at large scales is
one approach for investigating the role evolutionary biology plays in shaping
processes and patterns in the natural world. The results of this study indicate that
differences exist between phylogenetic diversity and species richness within
ecoregion sections, and suggest that phylogenetic diversity may be a more
meaningful measure of biodiversity because it accounts for evolutionary
relationships among the species in a community. Further, the analyses of
phylogenetic community structure detected intriguing regional patterns of
phylogenetic clustering and overdispersion. Additional research is necessary to
better understand the causes of these patterns, and to further explore the
correlations, at regional scales, between phylogenetic clustering and several
environmental variables.
The types of analyses described in this paper have potential for investigating
the evolutionary aspects of forest health. For example, these analyses could
assess whether regions with more phylogenetically clustered forest tree
communities are more susceptible to major environmental changes, because of the
possibility that a higher proportion of species will be affected by a given stressor.
Another important area of research using these methods could quantify the
evolutionary impacts of the loss of species, in terms of loss of ecological
functionality and evolutionary potential.
Finally, this study establishes that Forest Inventory and Analysis data are
suitable for addressing evolutionary biology questions at regional scales. In fact,
few if any other available data sets have the extent and resolution necessary to
conduct phylogenetic community analyses for forest communities at broad scales.
FIA inventory data have considerable potential for use in many such future
studies, including several with forest health applications. Future research, for
example, could incorporate FIA data to:
(1)
assess whether nonnative invasive forest tree species are more
phylogenetically related than expected by chance (Strauss and others
2006), which could allow for the identification of groups of closely
related nonnative species that may be likely to become invasive.
(2)
determine the phylogenetic community effects of the loss of species
or groups of species as a result of forest tree insects or pathogens,
such as chestnut blight, hemlock woolly adelgid, and white pine
blister rust.
14
USDA Forest Service Proceedings – RMRS-P-56
49.
(3)
test for phylogenetic signal in forest tree susceptibility to insect and
disease infestation, particularly from those with multiple hosts, and to
broad environmental changes such as climate change.
(4)
quantify the phylogenetic diversity and community structure of entire
forest plant communities, and of strata within those communities,
using FIA Phase 3 data.
Acknowledgments
I wish to thank Frank Koch, Mark Ambrose and Barb Conkling from the
Forest Health Monitoring Group at North Carolina State University for their
insights about this research and for their technical and administrative support. I
further thank Kurt Riitters from the U.S. Forest Service and Cam Webb of the
Arnold Arboretum at Harvard University for their technical support. This
research was supported in part through Research Joint Venture Agreement 06-JV11330146-124 between the USDA Forest Service and North Carolina State
University.
Literature Cited
Bailey, R.G. 1995. Description of the ecoregions of the United States. 2nd ed., rev. Misc.
Publ. No. 1391. Washington, D.C.: U.S. Department of Agriculture. 108 p. +
map.
Cavender-Bares, J.; Ackerly, D.D.; Baum, D.A.; Bazzaz, F.A. 2004. Phylogenetic
overdispersion in Floridian oak communities. American Naturalist. 163(6):823843.
Cleland, D.T.; Freeouf, J.A.; Keys, J.E. [and others], 2005. Ecological Subregions:
Sections and Subsections for the conterminous United States.[CD-ROM].
Washington, D.C.: United States Department of Agriculture, Forest Service.
Erwin, T.L. 1991. An evolutionary basis for conservation strategies. Science.
253(5021):750-752.
ESRI. 2006. ArcMap 9.2. Redlands, California: Environmental Systems Research
Institute Inc.
Faith, D.P. 1992. Conservation evaluation and phylogenetic diversity. Biological
Conservation. 61:1-10.
Forest Inventory and Analysis Program. 2007. The Forest Inventory and Analysis
Database: Database Description and Users Guide Version 3.0. Arlington,
Virginia: United States Department of Agriculture, Forest Service. 230 p.
Kembel, S.W.; Hubbell, S.P. 2006. The phylogenetic structure of a neotropical forest tree
community. Ecology. 87(7):S86-S99.
Miller, D.A.; White, R.A. 1998. A Conterminous United States Multi-Layer Soil
Characteristics Data Set for Regional Climate and Hydrology Modeling.
[Online].
http://www.soilinfo.psu.edu/index.cgi?soil_data&conus.
[Date
accessed: July 28, 2008].
Montréal Process Working Group. 2008. The Montréal Process, third edition. [Online].
http://www.rinya.maff.go.jp/mpci/rep-pub/1995/santiago_e.html. [Date accessed:
May 22, 2008].
15
USDA Forest Service Proceedings – RMRS-P-56
49.
Potter, K.M. in review. Evolutionary diversity and phylogenetic community structure of
forest trees across the conterminous United States. In: Potter, K. M.; Conkling, B.
L., ed. Forest Health Monitoring 2008 National Technical Report. Asheville,
North Carolina: United States Department of Agriculture, Forest Service,
Southern Research Station.
PRISM Group. 2008. 30-arcsec (800 m) gridded 30-year normal climate data (19712000). [Online]. http://www.prism.oregonstate.edu/products/matrix.phtml. [Date
accessed: July 30, 2008].
Rodrigues, A.S.L.; Gaston, K.J. 2002. Maximising phylogenetic diversity in the selection
of networks of conservation areas. Biological Conservation. 105(1):103-111.
Sechrest, W.; Brooks, T.M.; da Fonseca, G.A.B. [and others]. 2002. Hotspots and the
conservation of evolutionary history. Proceedings of the National Academy of
Sciences of the United States of America. 99(4):2067-2071.
Soltis, P.S.; Gitzendanner, M.A. 1999. Molecular systematics and the conservation of
rare species. Conservation Biology. 13(3):471-483.
Strauss, S.Y.; Webb, C.O.; Salamin, N. 2006. Exotic taxa less related to native species
are more invasive. Proceedings of the National Academy of Sciences of the
United States of America. 103(15):5841-5845.
Tofts, R.; Silvertown, J. 2000. A phylogenetic approach to community assembly from a
local species pool. Proceedings of the Royal Society of London Series BBiological Sciences. 267(1441):363-369.
United States Department of Agriculture Forest Service. 2008. National forest type data
development. [Online]. http://svinetfc4.fs.fed.us/rastergateway/forest_type/.
[Date accessed: May 13, 2008].
United States Geological Survey. 1996. GTOPO30 - Global Topographic Data. [Online].
http://edc.usgs.gov/products/elevation/gtopo30/gtopo30.html. [Date accessed:
July 30, 2008].
Webb, C.O. 2000. Exploring the phylogenetic structure of ecological communities: An
example for rain forest trees. American Naturalist. 156(2):145-155.
Webb, C.O.; Ackerly, D.D.; Kembel, S.W. 2007. Phylocom: Software for the Analysis of
Community Phylogenetic Structure and Character Evolution. Version: 3.41.
Webb, C.O.; Ackerly, D.D.; McPeek, M.A.; Donoghue, M.J. 2002. Phylogenies and
community ecology. Annual Review of Ecology and Systematics. 33:475-505.
Webb, C.O.; Gilbert, G.S.; Donoghue, M.J. 2006. Phylodiversity-dependent seedling
mortality, size structure, and disease in a Bornean rain forest. Ecology.
87(7):S123-S131.
Wikström, N.; Savolainen, V.; Chase, M.W. 2001. Evolution of the angiosperms:
Calibrating the family tree. Proceedings of the Royal Society of London Series
B-Biological Sciences. 268(1482):2211-2220.
16